An Efficient Data Aggregation Method in Wireless Sensor Network based on the SVD

The Wireless Sensor Networks (WSN), consists of thousand small and inexpensive sensing devices (Sensors) with a limited energy resources, memory, computational capability, and communication[1]. In WSN, most of the limited energy will be expended during the communication process, both reception and transmissions. Thus, reducing the amount of energy consumption during the transmission process can increase the WSN lifetime and overall efficiency. One of the most effective methods of the energy reduction in WSN is to reduce the amount of transmitted data between the sensor nodes and Cluster Heads(CH) as well as the transmitted data from cluster heads to the Base Station(BS) . These data reduction criteria can be accomplished by using data aggregation algorithms in which the data redundancy will be eliminated to reduce the amount of transmission data as well as to improve the energy consumption required for the transmission in WSN. During the data aggregation process, the collected data from the various sensor nodes and cluster heads will be combined to be processed by an efficient algorithm to reduce the possible data redundancy. The elimination of data redundancy will minimize the amount of useful data to be transmission and reduce the overall energy consumption which improves the efficiency and increase the life time of the WSN. The process of data reduction by eliminating the possible data redundancy to obtain the useful data with the minimum volume would be more effective when the data types are images. Our proposed algorithm will use Singular Value Decomposition (SVD) to extract the important features of the images needed to build the feature vectors. These feature vectors can be used to find the similar images, reduce the redundancy, and create useful images with least volume before they are transmitted to the cluster heads or from cluster to the base station. Our proposed algorithm will apply SVD on all images that are transmitted from sensor nodes to the CH and use the feature factors obtained from SVD analysis to eliminate the number of similar images in each cluster group and save the energy consumption due to reduction of the transmitted images from CH to the BS. Our algorithm is appropriate for both type of heterogeneous and homogeneous WSN regardless the type of algorithm that is used for the clustering process. The results of implementation show that the number of transferring images from CH to the BS is reduced about 50 percent in the proposed algorithm. In addition, the comparison of our proposed method with the LEACH algorithm, Figure 4, clearly indicated that our method has a better life time than the LEACH algorithm.